A civilisation's written record — translated, weighed, searched, and connected. At a standard scholars can cite.
Built and live in Abu Dhabi
Arabic is one of the world's great scholarly languages and the worst served by research AI. Machine translation flattens classical texts; nothing helps a researcher judge a source's credibility; the record sits disconnected. This platform was built to change that.
Publishable-grade Arabic→English translation of classical books — OCR, AI-vision proofread, meaning-preserving translation, full scholarly apparatus.
Academic viability assessment of any source — provenance, dating, citation integrity. Every grade evidence-backed; honestly deferred where evidence is thin.
The research assistant: a scholar's question returns recommended sources, a research plan, and a proposal draft — every claim verified against retrieved passages.
Historical family and tribal lineage reconstruction from archival sources, with per-fact provenance — from institutional pilots to commissioned family records.
Every output ships with its own audit: which claims were verified against retrieved corpus passages, which were not — flagged, never hidden — plus source coverage, citation density, and the exact compute cost. If the evidence is thin, the platform says so instead of guessing.
An AI that must cite — or confess. That is what makes its output citable, and what makes it defensible in front of a review committee.
Early-access pricing — billed by invoice during the pilot phase. Quotas and overage are enforced transparently in-product.
The platform is built and operated by researchers in the field it serves — classical Arabic historiography — and holds its outputs to an academic standard no general AI product attempts: full provenance, honest deferral, and per-claim verification at every layer.
An external academic-review model: published outputs are designed to carry independent scholarly sign-off. Honesty is engineered in at every layer — the platform flags what it cannot verify.